Code and my undegrad thesis (2015). In this thesis, I propose a different approach to dictionary learning called "non-sparse dictionary learning". My experiments show that my approach is faster, better than other sparse dictionary learning approaches.

Although people don't use dictionary learning any more today, I just want to publish here for reference purpose and it's fun to look back.

Accompanying paper:

@INPROCEEDINGS{7371815,
  author={Tuan, Nguyen Duc and Manh, Nguyen Quang and Sang, Dinh Viet and Huynh Thi Thanh Binh and Thuy, Nguyen Thi},
  booktitle={2015 Seventh International Conference on Knowledge and Systems Engineering (KSE)}, 
  title={A Study on Non-sparse Dictionary Learning for Pattern Classification}, 
  year={2015},
  volume={},
  number={},
  pages={371-376},
  doi={10.1109/KSE.2015.66}}

MIT